Anti-mutagenic Activity of DNA Damage-binding Proteins Mediated by Direct Inhibition of Translesion Replication*

DNA lesions that block replication can be bypassed in Escherichia coli by a special DNA synthesis process termed translesion replication. This process is mutagenic due to the miscoding nature of the DNA lesions. We report that the repair enzyme formamido-pyrimidine DNA glycosylase and the general DNA damage recognition protein UvrA each inhibit specifically translesion replication through an abasic site analog by purified DNA polymerases I and II, and DNA polymerase III (α subunit) from E. coli.In vivo experiments suggest that a similar inhibitory mechanism prevents at least 70% of the mutations caused by ultraviolet light DNA lesions in E. coli. These results suggest that DNA damage-binding proteins regulate mutagenesis by a novel mechanism that involves direct inhibition of translesion replication. This mechanism provides anti-mutagenic defense against DNA lesions that have escaped DNA repair.


DNA lesions that block replication can be bypassed in
Escherichia coli by a special DNA synthesis process termed translesion replication. This process is mutagenic due to the miscoding nature of the DNA lesions. We report that the repair enzyme formamido-pyrimidine DNA glycosylase and the general DNA damage recognition protein UvrA each inhibit specifically translesion replication through an abasic site analog by purified DNA polymerases I and II, and DNA polymerase III (␣ subunit) from E. coli. In vivo experiments suggest that a similar inhibitory mechanism prevents at least 70% of the mutations caused by ultraviolet light DNA lesions in E. coli. These results suggest that DNA damage-binding proteins regulate mutagenesis by a novel mechanism that involves direct inhibition of translesion replication. This mechanism provides anti-mutagenic defense against DNA lesions that have escaped DNA repair.
Mutations caused by DNA-damaging agents play a key role in cancer via activation of oncogenes and inactivation of tumor suppressor genes (1). DNA repair is the major defense mechanism of cells against DNA damage and its deleterious effects. The major repair strategy in both prokaryotes and eukaryotes is excision repair. It involves excision of the damaged region from DNA, followed by re-synthesis using the complementary undamaged strand as a template (2). Two excision repair mechanisms are known for DNA damage: nucleotide excision repair (NER), 1 which operates on a wide variety of DNA lesions (3), and base excision repair (BER) which is more limited in its range of substrate specificity (4). The key step in these mechanisms is the recognition of DNA damage. In the bacterium Escherichia coli, the protein UvrA is responsible for the recognition of a broad range of DNA lesions that are repaired by NER, whereas in BER a series of DNA glycosylases with a much narrower range of specificity recognize and excise damaged or foreign bases from DNA (2).
A DNA lesion that has escaped repair can cause a discontinuity in DNA in the form of a ssDNA region carrying the lesion. This occurs when the replication fork is blocked at the lesion (5,6), or occasionally because of complications in nucleotide excision repair (7). This gap/lesion structure poses a problem for excision repair; an attempt to eliminate the lesion will cause a double-strand break, which is highly lethal, but on the other hand, the gap must be filled in for DNA replication to be completed and cell division to occur.
Two tolerance mechanism that operate in E. coli are responsible for filling in the gap/lesion structures (2). A recombinational repair mechanism patches the gap via recombination using homologous sequences from the sister chromatid. This process does not remove the damaged base, but it enables an error-free bypass, resulting in the restoration of genome continuity. An alternative bypass mechanism, which is regulated by the SOS stress response, involves filling in of the gap by DNA synthesis. This translesion replication reaction is potentially mutagenic because of the miscoding properties of most DNA lesions (6,8). The process was termed SOS mutagenesis, or error-prone repair, because the gap is repaired (though the damaged nucleotide remains) with the concomitant formation of mutations. It is responsible for mutations caused by a large spectrum of DNA lesions, including UV lesions, bulky carcinogen adducts, and abasic sites, which share the property of being blocks to DNA replication (2).
Biological regulation is often achieved by a combination of both positive and negative affectors; however, so far little is known on negative affectors that act directly on error-prone repair (9,10). In this study, we examined the possibility that proteins that bind lesions on ssDNA act as inhibitors of mutagenesis by directly inhibiting translesion replication.

EXPERIMENTAL PROCEDURES
Materials-UvrA and RecA were purified as described (11)(12)(13). DNA polymerase I (6000 units/mg) was obtained from Boehringer Mannheim. DNA polymerase II, purified according to Ref. 14, was a gift from M. Goodman (University of Southern California, Los Angeles, CA), and DNA polymerase III (␣ subunit), purified as described (15), was a gift from M. O'Donnell (Rockefeller University, New York, NY). Fpg was a gift from J. Laval (Institut Gustave Roussy, Villejuif, France).
The 40-nucleotide-long templates AB1 (5Ј-GCT GTA CAA CGT CGT GAC TGF GAA AAC CCT GGC GTT ACC C-3Ј) and AB2 (5Ј-GGA AAA  CCC TGG CGT TAC CCF ACT TAA TCG CCT TGC AGC A-3Ј), each containing a site-specific abasic site analog (marked F), were synthesized and purified as described (16) and were gifts from M. Takeshita (State University of New York, Stony Brook, NY). Similar oligonucleotides with no damage were synthesized by the Synthesis Unit of The Biological Services Department of the Weizmann Institute of Science. The substrates were prepared each by annealing the complementary 32 P-end-labeled 17-mer primer to the 40-nucleotide-long synthetic DNA template as described (17). Essentially, all template molecules contain the abasic site analog (Ͼ99.98%), as deduced from the complete arrest of DNA synthesis at the lesion when bypass was assayed with pol I in the presence of 0.1 M KCl (Ͻ0.02% bypass, representing the limit of detection; Ref. 18).
Translesion Replication-The assay was performed essentially as described (17,18). The in vitro replication reaction mixture (25 or 50 l) contained 20 mM Tris⅐HCl, pH 7.5, 8 g/ml bovine serum albumin, 5 mM dithiothreitol, 0.1 mM EDTA, 4% glycerol, 10 mM MgCl 2 , 1 mM ATP and * This work was supported by grants from the Scheuer Research Foundation of the Israel Academy of Sciences, from the Israel Ministry of Science and the Arts, and from The Forchheimer Center for Molecular Genetics. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom all correspondence should be addressed.  20), which are minimal estimated concentrations that might be further increased by compartmentalization. Reactions were carried out at 30°C for the indicated time periods, after which they were treated with proteinase K and fractionated by electrophoresis on 20% polyacrylamide gels containing 8 M urea. End-labeled oligonucleotides were used as size standards. Samples were run at 1500 -1600 V for 3-4 h, after which they were visualized and quantified using a Fuji BAS 1000 phosphorimager.

UV Light and Methylmethanesulfonate (MMS) Mutagenesis-The
His reversion assay was performed essentially as described (10). Derivatives of E. coli AB1157 cells were UV-irradiated at 254 nm with the indicated doses using a germicidal lamp, after which they were plated on M9 minimal plates supplemented with 100 mg/liter each of Pro, Arg, Thr, Leu, Ile, and Val, and a limited amount of 2 mg/liter His. His ϩ colonies were counted after 3-4 days. The total number of cells was determined by plating the appropriate dilution of the culture on the same plate (the amount of His was sufficient to allow growth of several hundreds of colonies). The mutation frequency was calculated by dividing the number of mutants by the total number of cells. The assay for UV light-induced reversion of Arg auxotrophy was conducted according to the same procedure except that the concentration of His in the plates was raised to 100 mg/liter, and that of Arg was lowered to 3 mg/liter. MMS-induced His reversions were assayed as follows; cells were grown to OD 595 0.5, after which they were washed, resuspended in 10 mM Tris⅐HCl, pH 7.5, 0.15 M NaCl, and incubated for 30 min at 37°C with 0.5-2.0 g/ml MMS. His ϩ mutants were assayed as described above for UV mutagenesis.

RESULTS
Formamido-pyrimidine DNA Glycosylase Inhibits Translesion DNA Synthesis through an Abasic Site Analog-Abasic sites are a classical example of miscoding and replicationblocking DNA lesions, which cause mutations via the SOS response (22). Although polymerization frequently terminates at abasic sites, purified DNA polymerases were shown to be able to bypass abasic sites unassisted, usually incorporating a dAMP residue opposite the lesion (23)(24)(25)(26). In the current study, we utilized a model assay system that we have recently developed for translesion DNA synthesis, based on a synthetic DNA substrate containing a site-specific abasic site analog (17,18). The DNA substrate utilized consisted of a 5Ј-radiolabeled 17mer oligonucleotide primer, annealed to a 40-nucleotide-long synthetic ssDNA carrying a site-specific abasic site analog at position 21 (Fig. 1A). Replication of this substrate by a DNA polymerase resulted in extension of the radiolabeled primer, and this was assayed by urea-polyacrylamide gel electrophoresis, followed by quantitative measurements of the radiolabeled bands. Termination of DNA synthesis at the lesion is expected to yield products that are 19 or 20 nucleotides long, for termination prior to or opposite the lesion, respectively, whereas translesion replication is expected to yield a full-length 40nucleotide product (Fig. 1A).
We examined first the effect on bypass of formamido-pyrimidine DNA glycosylase (Fpg), a repair enzyme involved in BER of oxidative DNA damage. It catalyzes the removal of formamido-pyrimidines and 8-oxo-guanine from DNA, and it has an associated AP lyase activity, and thus is also an abasic sitebinding protein (27,28). The AP lyase activity leads usually to the cleavage of the phosphodiester bond; however, the lack of the hydroxyl group in the furanyl abasic site analog renders it resistant to breakage. The fact that Fpg binds but does not cleave the abasic site analog (29) makes it a convenient model system to assay inhibition of bypass.
We examined the effects of Fpg on bypass synthesis by each of the three known DNA polymerases of E. coli. We have shown previously that when the appropriate reaction conditions are used, each of the DNA polymerases can bypass the abasic site analog, unassisted by other proteins, at frequencies higher than reported previously (17). We used DNA polymerase I, DNA polymerase II, and the ␣ subunit of DNA polymerase III (pol III␣). In this bypass assay system, pol III core showed little bypass, and thus its ␣ subunit, which contains the polymerase active site, was used instead (17). As can be seen in Fig. 1, each of the DNA polymerases was capable of replicating through the abasic site analog, albeit at different rates. Addition of Fpg caused an inhibition of translesion replication by each of the DNA polymerases of up to 10 -20-fold (Fig. 1, B and C). This was evident from the decreased amount of bypass products, and from the accumulation of termination products at the lesion (Fig. 1B).
The UvrA Damage Recognition Protein Inhibits Bypass Synthesis-To examine whether other DNA damage-binding proteins inhibit translesion synthesis, we analyzed the effect of the UvrA protein on bypass of the abasic site analog. UvrA is the general DNA damage recognition protein of E. coli (2). It recognizes a large spectrum of DNA lesions (19), including abasic sites (30). When purified UvrA protein was added to the bypass reaction, it caused a strong 10 -20-fold inhibition of bypass synthesis by DNA polymerase I or by pol III␣, and it completely inhibited (900-fold) translesion synthesis by pol II (Fig. 2; Table  I). The termination of DNA synthesis in these experiments and in the experiments with Fpg was not caused by cleavage of the template by Fpg or by UvrA at the abasic site analog, because the abasic site analog is insensitive to the enzyme (29) and UvrA has no nuclease activity (2). This was confirmed by repeating the bypass reactions in the presence of Fpg or UvrA with a DNA substrate in which the 5Ј-end of the template strand was radiolabeled (data not shown). Thus, DNA damagebinding proteins of both the NER and BER pathways of repair strongly inhibit translesion replication.
The Inhibition of Bypass Synthesis by UvrA and Fpg Is Specific-The dependence of the inhibitory effect on the presence of the lesion in DNA was addressed by repeating the inhibition experiments with a control template with no damage, and the results are shown in Table I. DNA synthesis on the undamaged DNA by each of the three DNA polymerases was not affected by Fpg, compared with the 10 -20-fold inhibition on the damaged template. Thus, inhibition of bypass by Fpg was DNA damage-specific. The UvrA protein caused an inhibition of DNA synthesis on the undamaged template; however, its extent depended on the polymerase. DNA synthesis by pol II was inhibited only 2.7-fold, much weaker than the 900-fold inhibition of bypass synthesis. For pol I and pol III␣, inhibition by UvrA of DNA synthesis on the damaged template was 2-3-fold stronger than on the undamaged template (Table I).
To examine whether any protein that binds ssDNA inhibits bypass synthesis, we assayed the effect on translesion synthesis of RecA, a protein that binds ssDNA regardless of the presence of DNA damage (31). We found that RecA inhibited bypass synthesis by pol I and pol III␣ by 10% and 21%, respectively, and it stimulated bypass by pol II by 23%. Thus, RecA had only a marginal effect on bypass synthesis. This small effect was not due to lack of DNA binding, since under our conditions RecA caused a large retardation in the mobility of the template when assayed by the gel mobility shift assay (data not shown). Thus, not every ssDNA-binding protein inhibits translesion synthesis.
The UvrA Protein and DNA Photolyase Inhibit UV Mutagenesis in Vivo in the Absence of DNA Repair-The in vitro studies presented above raise the possibility that DNA damage-binding proteins have an anti-mutagenic activity separate from the one which is associated with the removal of DNA damage. If true in vivo, one would predict that SOS mutagenesis will be higher in a strain that lacks DNA damage-binding proteins when compared with a strain that contains these proteins. The examination of such a prediction should be made under conditions in which the repair capacity of the two strains is the same, and the only difference is the presence or absence of DNA damage-binding proteins. The mutagen that we examined was UV light. It was chosen because it provided a convenient experimental system (see below) in which the concept could be examined on DNA lesions other than abasic sites. UV light produces primarily cyclobutyl dimers and 6 -4 pyrimidinepyrimidone adducts (2). These lesions are chemically very different from abasic sites, and they are repaired by different error-free repair mechanisms. However, they give rise to mutations by the same SOS mutagenesis pathway, and thus they are functional homologues, as far as induced mutagenesis is concerned (6,22,32). In particular, the 6 -4 photoadducts share with abasic sites the properties of being both blocking and miscoding (33). The cyclobutyl photodimers are blocking, but only weakly miscoding (34,35).
Both types of UV lesions are repaired by the UvrABC NER system, although the 6 -4 adducts are repaired faster. The cyclobutyl dimers are also repaired by DNA photolyase, using light as a cofactor (2). By using a uvrC mutant (NER not operative) and working under yellow light (DNA photolyase not active), one creates a situation in which cyclobutyl dimers and 6 -4 adducts are not repaired. Under these conditions, the repair capacity of the uvrC strain is the same as a uvrA phr double mutant, except that in the uvrC strain both UvrA and DNA photolyase are still present and bind UV lesions (photolyase also binds cyclobutyl dimers in the dark; Ref. 36).
The prediction was that if UvrA and photolyase inhibit translesion replication in vivo, UV mutagenesis would be higher in the uvrA phr double mutant as compared with the uvrC mutant (Table II).
We compared UV mutagenesis and survival of an isogenic series of single uvrA and uvrC mutants, and the uvrA phr and uvrC phr double mutants. We used low UV doses, under which cell survival was high (70 -90%), conditions that are physiologically relevant. As can be seen in Fig. 3A, the frequency of UV light-induced His Ϫ 3 His ϩ reversions was 3.3-fold higher in  DNA, and on substrates containing a site-specific abasic site analog The reactions were carried out and analyzed as described in the legends to Figs. 1 and 2, except that undamaged templates AB1 and AB2 were used, each containing a guanine residue instead of the abasic site analog, and reaction times were 2, 4, and 6 min. The rates were calculated from the phosphorimager images of gels similar to those presented in Fig. 1 the uvrA phr strain, as compared with the uvrC strain. UV survival was identical, indicating that the repair capabilities of these strains were the same, as expected. Mutants carrying the uvrA or uvrC phr mutations exhibited intermediate mutagenicity. We also examined the same mutants in the Arg reversion mutagenesis assay (Fig. 3B). Similarly to the His reversions, UV-induced Arg Ϫ 3 Arg ϩ reversions were higher in the uvrA phr strain, as compared with the uvrC strain, although the effect was slightly lower (2.5-fold). These results indicate that, when present, UvrA and DNA photolyase eliminate as much as 70% of the UV-induced His ϩ mutations by a mechanism that does not involve DNA damage removal. Our model predicts that the anti-mutagenic effect of a particular DNA damage-binding protein will be directed only against the DNA lesions to which it binds. Thus, no antimutagenic effect of UvrA and DNA photolyase is predicted, for example, for alkylation damage in DNA. This prediction was examined by assaying MMS mutagenesis in the various mutants. Like UV mutagenesis, MMS mutagenesis is SOS-dependent. However, in contrast to UV light DNA damage, the alkylation DNA damage caused by MMS is repaired primarily by demethylation and by BER (2). Thus, it was expected that MMS mutagenesis would be the same in the uvr and phr mutants. As can be seen in Fig. 3C, there were no differences in MMS mutagenesis among the four strains, as expected, suggesting that the effect is specific for the appropriate combination of DNA damage-binding proteins and the DNA lesions to which they bind.
The most likely explanation for the lower UV mutagenesis in the uvrC mutant as compared with the uvrA phr mutant is binding of UvrA and DNA photolyase to UV lesions present on ssDNA regions, and physical obstruction of the bypass at-tempts by the polymerase, similar to our in vitro results with the abasic site analog. The magnitude of the inhibition of translesion DNA synthesis by DNA-binding proteins in vivo may be larger than the 3.3-fold effect observed in our UV mutagenesis experiments, because there may be other proteins that bind pre-mutagenic UV lesions in addition to UvrA and DNA photolyase. DISCUSSION This study addressed the possibility that DNA damage-binding proteins can intervene directly in the mutagenic bypass reaction and inhibit it, thus providing extra anti-mutagenic protection to the cell. Our results with the in vitro model system showed that the Fpg and UvrA repair proteins specifically inhibited in vitro translesion replication through an abasic site analog by purified DNA polymerases. These effects can be explained by the binding of the DNA damage-binding proteins to the damaged site in DNA, forming a physical obstacle to polymerization by the DNA polymerases. As the polymerase encounters the lesion, it pauses and readily dissociates from DNA (25,37,38). The presence of a DNA damage-binding protein may facilitate dissociation and/or inhibit rebinding of the polymerase to the primer-lesion site, leading to an inhibition of bypass synthesis. Such a mechanism could be common to many types of DNA lesions and the proteins that specifically bind them. In particular, replication of strongly blocking lesions is expected to be prone to this type of inhibition, because the DNA polymerases have difficulty in replicating them in the first place, strongly favoring dissociation over bypass (6).
UV lesions and abasic sites give rise to mutations by a similar translesion replication mechanism (6,8,22). Our findings that, in the absence of DNA repair, UV mutagenesis was higher in a strain lacking UvrA and DNA photolyase and that Fpg and UvrA inhibit bypass synthesis in vitro can be explained by the same basic molecular mechanism. According to this mechanism, a DNA lesion that has escaped repair and is located in a ssDNA region is recognized by both specific and general DNA damage-binding proteins. These proteins are unable to eliminate the lesion by excision repair, because that would cause a lethal double strand break. Instead, the binding of these proteins to the lesion inhibits bypass synthesis, and thus causes a reduction in mutagenesis.  Consistent with our suggestion is the observation that overproduction of DNA photolyase reduced in vivo UV mutagenesis (39). A related observation was the report that mutations due to a site-specific O 6 -butylguanine in phage x174 increased 8-fold in a uvrA6 mutant. However, the interpretation of this experiment is complicated by the fact that a missense uvrA6 mutant, rather than a deletion mutant, was used, and because repair mechanisms responsible for removal of alkylation damage were not neutralized in the cell (40).
The exact events that occur in a cell following UV irradiation are not fully understood. However, it is clear that ssDNA regions are formed due to the inhibition of DNA replication by blocking UV lesions that were not repaired (5,6). These regions may serve as binding sites for DNA damage-binding proteins, thus suppressing the error-prone translesion replication. The final mutagenicity is the outcome of the activity of factors that facilitate UV mutagenesis, such as RecA, UmuD, and UmuC, and factors that inhibit the process, such as UvrA and DNA photolyase. How does the binding of UvrA and photolyase lead to an anti-mutagenic effect? One possibility is that this intermediate cannot be repaired and will lead to cell death. This would have predicted differences in UV survival of uvrA phr and uvrC mutants; however, such a difference was not observed. We favor the idea that, while inhibitory to the mutagenic translesion DNA synthesis, the binding of DNA damage-binding proteins does not interfere with error-free recombinational repair. This enables continuation of chromosome replication and cell division, without creating a mutation.
In summary, our results suggest that the mutagenic effects caused by agents such as UV light, which produce replication blocks, are suppressed by DNA damage-binding proteins. This provides a second line of defense, which reduces the mutagenic effects of DNA lesions that have escaped DNA repair. It operates by the direct inhibition of translesion replication, mediated via binding of DNA damage-binding proteins to the premutagenic ssDNA regions carrying the DNA lesions. This represents a new regulatory mechanism of induced mutagenesis, and a new anti-mutagenic function for DNA damage-binding proteins. At this point, we do not know whether a similar mechanism operates in eukaryotes. However, based on the great similarity in basic DNA repair mechanisms between E. coli and eukaryotes, direct inhibition of translesion DNA synthesis by DNA damage-binding proteins, such as the human XPA or the yeast Rad14 proteins, is likely to be involved in the regulation of induced mutagenesis in eukaryotes as well.